Grid-based Map Analysis Techniques and GIS Modeling


GIS in the Rockies Conference, Denver    October 2004


Joseph K. Berry


Keck Scholar in Geosciences, University of Denver

Principal, Berry & Associates // Spatial Information Systems

2000 South College Avenue, Suite 300, Fort Collins, Colorado  80525

Email — Web


Situation:  Most desktop mapping and GIS applications have focused on mapping and spatial data management responding to inventory assessments of "Where Is What" involving digital maps and linked databases (geo-query of mapped data).  GIS modeling provides new analytical tools and processing structures for incorporating spatial relationships that address "Why and So What" within a decision-making context (map analysis and modeling).  This extension of descriptive to prescriptive mapping involves entirely new spatial reasoning concepts and skills that are not reflected in our paper map legacy.


Description:  This intermediate level workshop discusses and demonstrates several techniques for spatial analysis and data mining using numerous examples from natural resources management, geo-business and precision agriculture.  The discussion focuses on concepts, procedures and practical considerations in successfully applying grid-based map analysis in GIS modeling.  The material presented encapsulates numerous “Beyond Mapping” columns by the instructor published in GeoWorld magazine and compiled into the online book Map Analysis (, select Map Analysis).


Who Should Attend:  GIS professionals who are currently involved in the development of systems that analyze spatial data should attend.  Prior GIS exposure and familiarity with basic statistical concepts are recommended.


Joseph K. Berry is a leading consultant and educator in the application of Geographic Information Systems (GIS) technology.  He is the principal of BASIS, consultants and software developers in GIS technology and the author of the Beyond Mapping column for GeoWorld magazine for over twelve years.  He has written over two hundred papers on the theory and application of map analysis techniques, and is the author of the popular books Beyond Mapping and Spatial Reasoning.  Since 1976, he has presented college courses and professional workshops on geospatial technology to thousands of individuals from a wide variety of disciplines.  Dr. Berry conducted basic research and taught courses in GIS for twelve years at Yale University's Graduate School of Forestry and Environmental Studies, and is currently the W. M. Keck Visiting Scholar in Geosciences at the University of Denver and a Special Faculty member at Colorado State University.  He holds a B.S. degree in forestry from the University of California, Berkeley, a M.S. in business management and a Ph.D. emphasizing remote sensing and land use planning from Colorado State University.





Workshop Schedule


                                 (Note: times are approximate)





Maps as Data

¾    Discrete map objects vs. continuous geographic space

¾    Grid data types, structures and display


9:00- 9:30am

Surface Modeling

¾    Point density analysis

¾    Spatial interpolation

¾    Map comparison



Spatial Data Mining

¾    Linking geographic and data space

¾    Map similarity

¾    Clustering mapped data

¾    Map regression

¾    Future geo-statistical tools


10:00- 10:30am


10:30- 11:00am

Spatial Analysis

¾    Fundamental classes of analytical operations

¾    Suitability mapping

¾    Measuring effective distance/connectivity


11:00- 11:30am

Spatial Analysis (continued)

¾    Visual exposure analysis

¾    Analyzing landscape structure


11:30- 12:00am

GIS Modeling

¾    Modeling structure

¾    Processing hierarchy and analysis levels

¾    Calibrating and weighting model criteria

¾    Simulating alternative scenarios and perspectives



Workshop Notes


Joseph K. Berry, email, website


MapCalc Applications at

Map Analysis Book at


Cartography– manual map drafting (paper map legacy for thousands of years)

Computer Mapping– automates the cartographic process (70s)

Spatial Database Management– links computer mapping techniques with traditional database capabilities (80s)


Map Analysis and GIS Modeling– representation of relationships within and among mapped data (90s)…

ü      Surface Modeling– maps the spatial distribution of a set of point sampled data,

ü      Spatial Data Mining– characterizes the “numerical” relationships among mapped data and develops predictive models,

ü      Spatial Analysis– derives new information based on “contextual” relationships among mapped data,

ü      GIS Modeling– logical processing of spatial information to characterize a system or solve a problem.

(See Map Analysis, “Topic 4” for more information)


Raster refers to image display (map values represent the color assigned to each dot; e.g., scanned topographic maps–DRGs or aerial photos–DOQs) while Grid refers to map analysis (map values have all of the rights, privileges and responsibilities of a map-ematics).

Grid Data Structure the Analysis Frame provides consistent “parceling” needed for map analysis and extends points, lines and areas to Map Surfaces.

(See MapCalc Applications, “Short Video Demos” for more information)  


Shading Manager options include # of Ranges, Calculation Method (e.g., Equal Ranges with same range for each interval and Equal Count with same number of cells for each interval) and Color Pallet/Ramp selection.

Grid Display Types are Lattice that forms a smooth “wireframe” by connecting cell centroids with lines whose lengths are a function of elevation differences and Grid that forms extruded grids whose heights are a function of elevation differences.

Grid Data Types are characterized by their Numeric Distribution (independent integers versus range of values) and their Geographic Distribution (abrupt boundaries versus gradient).  A Discrete map has values that simply represent categories (e.g., a covertype map) that form sharp abrupt boundaries) whereas a Continuous map has values that represent a spatial gradient (e.g., a slope map). 

(See Map Analysis, “Topic 18” for more information)

(See MapCalc Applications, “Display Types” and “Data Types” for more information)


Surface Modeling maps the spatial distribution and pattern of point data…

ü      Map Generalization– characterizes spatial trends (e.g., titled plane) by considering all of the samples at once as it fits a surface,

ü      Spatial Interpolation– derives spatial distributions (e.g., IDW, Krig) by considering small, localized set of samples throughout the map area (roving window), and

ü      Other– roving window/facets (e.g., density surface; tessellation)

(See Map Analysis, “Topics 2 and 8” for more information)


Data Mining investigates the “numerical” relationships in mapped data…

ü      Descriptive– calculates aggregate statistics (e.g., average/stdev, similarity, clustering) that summarize mapped data,

ü      Predictive– develops relationships among maps (e.g., regression) that can be used to forecast characteristics or conditions at other locations or times, and

ü      Prescriptive– uses descriptive and predictive information to optimize appropriate actions.

(See Map Analysis, “Topics 7 and 16” for more information)


Spatial Analysis investigates the “contextual” relationships in mapped data…

ü      Reclassifying Maps– New map values are a function of the values on a single existing map… no new spatial information is created,

ü      Overlaying Maps– New map values are a function of the values on two or more existing maps… new spatial information is created,

ü      Measuring Distance– New map values are a function of the simple or weighted distance or connectivity among map features, and

ü      Summarizing Neighbors– New map values are a function of the values within the vicinity of a location on an existing map.

(See Map Analysis, “Topic 22” for more information)


Measuring Distance– the concept of Distance as the “shortest straight line between two points” is expanded to Proximity by relaxing the assumption of only “two points” then expanded to Movement by relaxing the assumption of “straight-line” connectivity.

(See Map Analysis book, “Topics 13, 14, 17, 19 and 20” for more information)

(See MapCalc Applications, “Determining Proximity” and “Creating an Up-Hill Road Buffer”)


Calculating Visual Exposure– a Viewshed identifies all locations that can be seen from a view point(s) while Visual Exposure develops a relative scale indicating the number of times each location is seen from a set of viewer points, such as a road network.

(See Map Analysis book, “Topic 15” for more information)

(See MapCalc Applications, “Determining Visual Exposure” and “Modeling Visual Exposure”)” for more information)


Summarizing Neighbors– a Diversity Map indicates how many different types, a Roughness Map identifies the variation in slope values, and a Density Map reports the total value within a specified distance of each grid location.

(See Map Analysis, “Topic 9” for more information)

(See MapCalc Applications, “Assessing Covertype Diversity”)” for


GIS Modeling Procedures– There are three basic types of GIS Models

ü      Statistical Models– based on numerical relationships (e.g., crop yield),

ü      Process Models– based on physical (e.g., erosion potential), and

ü      Suitability Models– based on logically sequenced decision criteria similar to a recipe (e.g, animal habitat)…

  • Binary Model– identifies areas that are acceptable based on combining binary maps (0 and 1), 
  • Ranking Model– develops a ranking of areas based on the number of criteria that are acceptable (0 to 3), and
  • Rating Model– develops a “goodness” scale (0 to 9 best) and calculates the average rating for each grid cell.

(See Map Analysis, “Topic 23” for more information)




    MapCalc LearnerTMStudent Tutorial Version (CD) with MapCalc and Surfer Tutorial systems, Exercises/databases, application demos and text; 100x100 configuration; single seat license for educational use only; US$21.95 plus shipping.

(See for more information)